The use of ultrashort light pulses to study coherent interactions in atomic and molecular systems has advanced rapidly in recent years. One exciting area of current research is in ‘‘coherent control,’’ where light pulses that have been precisely shaped in amplitude and phase can selectively ‘‘drive’’ @1# a chemical reaction @2#, molecular vibration @3#, or other process such as nonlinear-optical conversion of light into the extreme ultraviolet region @4#. However, current experimental techniques are still very limited in their ability to coherently control the evolution of quantum systems. In the most general case, coherent control techniques require phase-coherent, femtosecond, temporally shaped pulses over a large region of the spectrum, to be able to access all possible intermediate states in a multistep quantum ‘‘pathway’’ to achieve the desired outcome @5#. Broadest-bandwidth ultrashort-pulse lasers generate only a fractional bandwidth on order of 30% of their carrier frequency. Nonlinear-optical techniques such as white-light continuum generation @6# and parametric amplification @7# can be used to generate coherent light over a broad spectrum. However, these techniques often suffer from poor efficiency. Furthermore, often the quantum transitions for a process of interest for coherent control are concentrated in a few disparate regions of the spectrum. It would thus be desirable to be able to take two separate laser systems, generating light with distinct optical properties, and precisely synchronize the output of both lasers, essentially generating a single, composite coherent light field. This ability to precisely synchronize separate, pulsed laser sources is an important step toward the ultimate, ‘‘arbitrary light wave-form generator.’’ It is also important for a number of other technologies, such as midinfrared light generation through difference frequency mixing @8#, for experiments requiring synchronized laser light and x rays or electron beams from synchrotrons @9#, and also for the synthesis of light pulses with shorter duration than is obtainable from any single individual laser @10#. Synchronization and phase locking of separate femtosecond lasers will also have a strong impact in precision frequency metrology based on optical frequency combs @11,12#, providing the capability of a wide-bandwidth phase-coherent frequency network covering various spectral regions. Frequency-domain control of femtosecond combs not only provides an effective means to transfer the stability of a cw optical oscillator to the entire comb @13#, but also has a strong impact on the time-domain evolution of carrierenvelope phase @14#. The frequency mode spacing of the femtosecond comb is equal to the inverse of the cavity round trip time of the laser; hence reduction of the phase noise of the repetition rate spectrum directly improves the control of the timing jitter. To date, previous work in synchronizing separate modelocked Ti:sapphire lasers demonstrated a timing jitter of at best a few hundred femtoseconds @15,16#. Since it is now routine to generate pulses with duration ,20 fs, improved techniques would make it possible to take full advantage of this time resolution for immediate application in various fields. In this Rapid Communication, we demonstrate robust synchronization of pulse trains from two separate femtosecond lasers, with a timing jitter of ,5 fs, at a bandwidth of 160 Hz, observed over several minutes. The two independent mode-locked Ti:sapphire lasers @17# operate at 780 and 820 nm, respectively, with ;100-MHz repetition rates. The bandwidth of both lasers corresponds to a transform limit of ,20 fs mode-locked pulses. Our synchronization scheme, shown in Fig. 1, employs two high-speed photodiodes to detect the two pulse trains. These signals are then input to four phase locked loops ~PLL’s ! working at different timing resolutions. Using the first PLL, we synchronize the repetition rate of